With the growth of cloud computing and multi-tenant architectures, visibility of application health, such as trends in errors experienced by the application over a network, has become an important feature to service providers and consumers alike. Past research in network health largely focused on the observation of uncorrectable errors experienced by applications, such as dropped packets, checksum errors, and parity errors. Armed with this observed data, network hardware can be designed with defensive techniques such as error-correcting code (ECC) and forward error correction (FEC) to prevent uncorrectable errors. Furthermore, network applications can be written to react and recover should an uncorrectable error occur. However, these solutions fail to capture, analyze and score correctable network errors to provide visibility of network-wide application health and alerts to applications and users before a catastrophic failure occurs.